The aim of this paper is first to review recent developments in the domain of musical information retrieval and then to present some methods developed at the Sound and Vision Engineering Department of the Gdansk University of Technology, Poland. Especially important for music retrieval systems is to find optimum music representation. This is can be done using the so-called FED decomposition first. This algorithm is also used for musical duet separation. In this context the evaluation of the efficiency of the FED algorithm based on the ANNs is given. Results of the performed experiments are shown and conclusions are derived.
展开▼